In response to a new European Union law, AI platforms are implementing new schemes for watermarking the content they generate. Anthropic recently disclosed its future Claude models will use SynthID-Text, an approach Google created and released as open source. It uses a secret key that subtly changes the process a model uses for choosing the next word in a sentence. Whereas a top next word choice might be “cloudy,” the key might change it to “overcast.” Anyone who knows the key can determine if it was generated by the platform using it.

New research shows that SynthID-Text can change not just word selection but also the tools a model invokes and the chances it will adhere to or disregard safety guardrails it has been trained to follow. The threat can become greater in the face of an adversarial prompt, in which an attacker attempts to cause a model to carry out a harmful action, such as revealing a password or other sensitive information. Instructions that normally wouldn’t be followed will, in some cases, be performed once the watermarking is deployed. The finding underscores the need for developers to thoroughly test how their LLMs and agents behave when watermarking is in place.

Changing safety behavior

“As compared to the same models without watermarking, it is definitely going to change their behavior, especially when we place it under adversarial conditions, or we make these models call tools when they’re powering an agent,” Andrea Siposova, an AI security researcher at Lasso Security, told Ars. “Watermarking is made to not be perceptible to a reader, but we know that when we are changing anything about what the model is generating, it is going to cause some tradeoffs, it’s going to show up somewhere.”

Read full article

Comments

© Getty Images

The International Meteor Organization, the nonprofit that coordinates and publishes amateur and professional observations of meteor phenomena, said its infrastructure has suffered a “critical blow” from a cyberattack.

“We recently suffered a cyberattack that dealt a critical blow to aging infrastructure, taking much of our site offline," a static page on its website on Wednesday said. “We expect several weeks of partial downtime as we transition to new infrastructure and services.”

“I am very sad to see the site down”

In the meantime, the IMO said it’s prioritizing the reporting of fireball observations, which can be reported here. The organization is also providing some information on its Facebook page.

Read full article

Comments

© Interntional Meteor Organization

In record time, Meta’s camera-adorned smart glasses have turned from a lucrative and well-reviewed consumer product into a widely-reviled and privacy-infringing tool of abuse in the arsenal of creepy men.

The backlash has grown so much that many owners of the “pervert glasses” have become too afraid to use them in public, a PR disaster for what once was an accessory that was flying off the shelves.

Meta is evidently alarmed by the growing criticism, desperately trying to implement a “kill switch” and remotely cutting off “manfluencers” by bricking their glasses.

And now, as The Information reports, the Mark Zuckerberg-led company is now looking to release a camera-free pair as well later this year — the clearest indication yet that Meta is losing in the court of public opinion.

Whether removing the camera from the glasses will be enough to assuage ongoing concerns over privacy remains to be seen. The camera-free glasses, code-named Luna, will still feature an array of six microphones to allow users to speak to the company’s AI chatbot and its recently-released AI assistant, called Muse. They’ll presumably pick up audio from users’ environments as well, where bystanders may not be aware of the eavesdropping.

The news comes just days after two families filed a proposed class action lawsuit, alleging Meta was extracting biometric information from people’s Instagram and Facebook photos to build a face-recognition system for its smart glasses.

The glasses could be the ultimate litmus test on what drew buyers to the pair of spectacles in the first place. By relying on microphones alone, Meta is betting that people are willing to shell out hundreds of dollars just to talk to its flawed AI companions — not take first-person view photos.

Project Luna could be unveiled as soon as next week’s Meta Connect conference. It’ll be interesting to see how the company will address the major backlash to its camera-equipped smart glasses, if at all. An admission that its consumer product is facilitating some incredibly creepy behavior in public feels unlikely — but it’ll certainly be the elephant in the room.

More on the glasses: Meta Is Using Instagram Users’ Photos to Build a Universal Facial Recognition System for Its Hated Smart Glasses, Class Action Lawsuit Claims

The post Zuckerberg’s Pervert Glasses Have Been So Publicly Shamed That Meta’s Reportedly Now Working on a Version With No Camera appeared first on Futurism.